Field sampling and processing
For each tree species, four healthy individuals with a similar diameter at breast height were randomly selected. To ensure sampling independence, the selected individuals were at least 10 m apart. We sampled the green and senesced foliage of these trees in autumn 2018. Specifically, the fully mature and expanded foliage from the most distal branches of each tree was cut from the sun-exposed canopy using a tree trimmer and bamboo extensions. Green foliage with visible biological or physical damage was excluded. More than 60 green leaves from each species were selected for further analysis. Thirty leaves per tree were placed in a valve bag and stored at 4 °C to measure morphological traits. The remaining samples were placed in envelopes to measure chemical traits. Due to the tree height and sampling difficulties, we spent two weeks per month during the whole autumn collecting senesced foliage. The senesced foliage was sampled directly beneath the focal trees from the recently fallen fresh litter on the ground (Huang et al. 2007). The senesced foliage with no visible degradation was well mixed for each species. The first set of subsamples was air-dried for chemical analyses, and the second set of subsamples was used for litter decomposition experiments.
Belowground samples were collected when sampling aboveground foliage. Four 1 × 1 m plots were identified within a 2 m distance from the tree stem in four directions from each target tree. The surface mineral soils (0–20 cm) at the plots were loosened with a pickaxe and spade, and sieved (2-mm mesh size) to sample the initial materials. Meanwhile, intact roots (more than five root orders) were cut from the main lateral roots, following the procedure described by Guo et al. (2008). Each root sample was divided into two subsamples for different analyses. One subsample was gently washed in deionised water to remove adhering organic matter and immediately fixed in formalin-aceto-alcohol solution (90 ml 50% ethanol, 5 ml 100% glacial acetic acid, and 5 ml 37% methanol) to measure anatomical traits. The other subsample was placed into valve bags, incubated with ice bags, and stored at -20 °C for later dissection and analysis of morphological and chemical traits (Guo et al. 2008).